Study On Spectral Characteristics And Estimation Model Of Urban Plant Leaf Surface Dust

Urban environmental problems are getting more and more attention.Atmospheric pollution,especially dust pollution,is one of the major environmental pollution.Atmospheric particulates not only affect the environment,but also have serious harm to human health.The plant can retain and adsorb atmospheric particulates,and the study of the dust ability of plants has important reference value and significance to the detection of urban air pollution.In this paper,the influence of leaf dust on the spectral characteristics of plants was explored by measuring the dust and spectral data of common greening tree species in xuhui district and minhang district of Shanghai.In order to establish the estimation model between the dust emission and the plant reflection spectrum,the spectral bands that are sensitive to the dust retention were selected.The effects of water and chlorophyll on the estimation model of dust retention were analyzed.This study can be used as the basis for estimating the dust retention in the city directly using remote sensing data in the future.The main results of this paper are as follows:(1)Dust has an effect on the reflection spectrum of plants.The spectral reflectance of different tree species changed with the increase of the amount of dust in the visible light(400~650nm).There is a consistent pattern in the near infrared band(750~1350nm).The spectral reflectance of the plant decreases with the increase of the dust-retention amount,that is,the spectral reflectance value of dust-free blades is greater than that of dusty blades.The difference between the reflectance of dusty and dustless leaves of Camellia is the largest.Dust retention does not change the “three-sided position” of the plant and does not cause “red edge movement”,but it will reduce the difference between “main peak” and “secondary peak”.Dust retention has an inhibitory effect on the "three-sided amplitude" and "three-sided area" of plants.The "triangular value" and "trilateral area" of dusty blades are larger than dustless blades.The effect of dust retention on the spectral reflectance of the canopy is the same as that of the leaves.(2)Combine the dust retention with spectral bands and spectral indices for correlation analysis,so as to select the bands and indices that are more sensitive to dust-retention.The spectral bands mainly include the original spectrum,logarithmic spectrum,first derivative spectrum and second derivative spectrum.Studies showed that the sensitive spectral regions and spectral segments of different tree species in the same spectral transformation form are closer.Green band(500~600nm),red band(620~760nm),near-infrared band around 1100 nm and the reflection peaks in the middle far infrared(1600~1800nm)are more sensitive areas.The spectral index was divided into existing spectral index and structured spectral index.According to the analysis of existing spectral indices,among the six tree species,the spectral indexes with strong correlation between Viburnum odoratissimum Ker-Gawl and Photinia serrulata were basically the same,and the genus of gold leaves and Camellia were basically the same.The correlation coefficients between the dust-retention capacity and the corresponding spectral index of the six tree species were not very high,and the water index and plant senescence reflectance index were relatively high correlation indexes.For structural spectral index analysis,the normalized index has a higher correlation,followed by the difference index,and the ratio index.The dust capturing capacity and the spectral index of the structure are highly correlated.(3)Based on the selected sensitive spectral bands and indices,multiple linear regression models and partial least-squares regression models were used to estimate the dust retention.In the estimation model of dust capturing capacity,the Viburnum odoratissimum Ker-Gawl is better,and the worst is the gold-leaf privet.The second derivative of the four spectrum indices,original spectrum,first derivative,second derivative and pseudo-absorption coefficient,has better prediction ability for leaf dust.The estimation model of dust capturing capacities based on the existing spectral index has lower precision,and the estimation model of dust capturing capacities based on the spectral construction index has a good effect.For the spectral bands,the partial least-squares regression model is better than the multiple regression model.There are no advantages or disadvantages for the spectral index.Multiple regression can be used when the number of independent variables is small,and partial least-squares regression can be used when the number of independent variables is large.(4)Considering the influence of moisture and chlorophyll on the estimation model of dust capturing capacities,analyzed the differences of plant spectral curves and their correlation with spectral bands and indices under different water conditions and pigment concentrations when there is dust and no dust.The study showed that the spectral curves of different water retention in the visible light range are not affected by the dust.The dust in the 750~1350nm band has a great influence on the spectral curve with less water retention,which is reflected in the decrease of the reflectivity.The accuracy of the model for estimating the dust susceptibility of Viburnum odoratissimum Ker-Gawl,Photinia serrulata and Euonymus japonicus cv.Aureo-ma was decreased with the increase of the water level,and the accuracy of the model constructed under the condition of less water was the highest.Dust retention does not change the order of reflectance spectra of different color grades at 400 to 700 nm.The dust in the range of 700-1100 nm has a significant influence on the spectrum with high pigment concentration.The moderate concentration of plant pigments is conducive to the construction of a better model for the estimation of dust retention.